AIMD-95-135 Medicare Claims: Commercial Technology Could ... · Technology Could Save Billions Lost...

United States General Accounting Office

GAO Report to Congressional Requesters

May 1995 MEDICARE CLAIMS

CommercialTechnology Could SaveBillions Lost to BillingAbuse

GAO/AIMD-95-135

GAO United States

General Accounting Office

Washington, D.C. 20548

Accounting and Information

Management Division

B-261034

May 5, 1995

The Honorable Pete V. DomeniciChairmanThe Honorable James ExonRanking Minority MemberCommittee on the BudgetUnited States Senate

The Honorable Tom HarkinRanking Minority MemberSubcommittee on Labor, Health and Human Services, Education, and Related AgenciesCommittee on AppropriationsUnited States Senate

Medicare is the nation’s largest health insurer, serving almost one in everyseven Americans. The Medicare program cost $158 billion during fiscalyear 1994 and is expected to rise to $286 billion by 2000. Federal outlaysfor physician services and supplies—one category of Medicarespending—totaled almost $36 billion in 1994. Given the magnitude of theseoutlays in a time of budgetary constraint, it is increasingly important toensure that program funds are not lost to fraud, waste, or abuse. As werecently reported, the Medicare program is plagued by billing abuse due toinadequate funding for fraud and abuse prevention activities, unevenimplementation of payment controls, and flawed payment policies.1

Avoiding these preventable losses would help control Medicare costswithout affecting beneficiary services or provider fees.

The Department of Health and Human Services’ (HHS) Health CareFinancing Administration (HCFA)—the agency responsible foradministering Medicare—contracts with 32 insurance companies, calledcarriers, to process and pay claims for physician services and supplies. Akey payment control these contractors use to prevent losses from fraud,waste, and abuse is claims processing computer systems that reviewclaims before payment is authorized. One type of abuse these systemsdetect is called code manipulation; this occurs when providers submitclaims containing an inappropriate combination of billing codes that can,if not detected and corrected, lead to overpayment for the servicesprovided. Many private and some public insurers, following health

11995 High-Risk Series: Medicare Claims (GAO/HR-95-8, February 5, 1995).

GAO/AIMD-95-135 Medicare Claims: Technology Could Save BillionsPage 1

B-261034

insurance industry best practices, use specialized commercial computersystems to detect these billing code abuses.2

This report responds to your request that we determine whether HCFA

should use commercial systems to detect code manipulation rather thancontinuing to develop its own capabilities in this area.3 Our objectiveswere to (1) determine whether commercially available codemanipulation-detection systems can reduce Medicare costs, (2) evaluatewhether HCFA’s development approach is likely to generate savingscomparable to that possible with commercial systems, and (3) assesswhether commercial systems are cost effective.

Results in Brief Based on a test in which four commercial firms reprocessed samples ofover 200,000 paid Medicare claims, we estimate that commercial codemanipulation-detection systems could have reduced federal outlays forphysician services and supplies, on average, by $603 million in 1993 and$640 million in 1994. This represents about 1.8 percent of Medicarepayments for such services and supplies, which is consistent with theactual savings achieved by private and public insurers that use commercialsystems. Also, because beneficiaries are responsible for about 22 percentof the HCFA-authorized payment amount (in the form of deductibles andcopayments), we estimate that they could have saved $134 million in 1993and $142 million in 1994. The test results also indicate that only a smallproportion of providers are responsible for most of the abuse: less than10 percent of providers in the sample had a miscoded claim.

HCFA is enhancing its ability to detect code manipulation, however, ouranalysis shows that its efforts will not match commercial systemcapabilities or savings. One reason is that HCFA’s approach does notaddress the types of abuse that accounted for about one-third of the lossescommercial systems identified. In addition, the types of abuse that arebeing addressed will not be fully prevented. Because commercial firmsspecialize in developing computer systems to detect billing abuse, they arebetter equipped than individual insurers to develop effective codemanipulation-detection capabilities. According to commercial firmofficials, the cost to implement and operate commercial systems for 1 year

2These specialized systems supplement rather than replace claims processing systems that performother important functions, such as determining whether the patient is entitled to Medicare benefits andcalculating deductible and coinsurance amounts.

3This report does not address other types of abuse, such as billing for inappropriate, unnecessary, orexcessive services.


B-261034

would range between $10 million and $20 million for all 32 Medicarecarriers.

Background Authorized in 1965 under title XVIII of the Social Security Act, Medicareprovides health insurance for about 33 million elderly people and about4 million others with disabilities or end-stage renal disease. In fiscal year1994, HCFA paid about $100 billion for inpatient, home health, and skillednursing care, and about $57 billion for noninstitutional care.Noninstitutional care covers physician services and supplies ($36 billion);and services at hospital outpatient facilities ($13 billion), group practices($5 billion), independent laboratories ($2 billion), and some home healthagencies ($120 million). Noninstitutional costs have increased morerapidly than inpatient hospital costs over the past decade, as health careservices shifted from primarily an inpatient setting to outpatient andphysician’s office settings.

Physician Services andSupplies Vulnerable toCode Manipulation

Code manipulation is a problem that is faced by all health insurers.Medicare pays health care providers a fee for each covered medicalservice provided to eligible beneficiaries. Each service is identified usingthe American Medical Association’s uniformly accepted coding system,called the Physicians’ Current Procedural Terminology (CPT). Medicareand most private insurers have developed or license fee schedules that useCPT codes and their accompanying narrative descriptions as the basis forpaying providers.

However, because the coding system is complicated, providers andinsurers often have difficulty identifying the codes that most accuratelydescribe the services provided. The coding system is difficult to usebecause it attempts to identify codes for all accepted medical procedures,including codes to describe minor procedures that are components ofmore comprehensive procedures. Payment policies add to the difficulty.For example, the fee for surgery often includes the cost of related servicesfor the global service period, that is, for a set number of days before andafter the surgery. To prevent overpayment in these cases, insurers need toidentify when claims for surgery include codes that represent relatedservices and reduce the payment accordingly. It is also difficult forproviders and insurers to maintain proficiency in proper coding practicesbecause a substantial number of the codes are changed each year.


B-261034

These complexities can inadvertently lead providers to submit improperlycoded claims. They also make insurers vulnerable to abuse from providersor billing services that4 attempt to maximize reimbursements byintentionally submitting claims containing inappropriate combinations ofcodes. HCFA has implemented and communicated policies that prohibitcommon abuses such as unbundling, global service period violations,duplicate procedures, and inappropriate use of assistant surgeons. Table 1defines these categories of abuse.

Table 1: Categories of AbuseCategory Description

Unbundling Billing for two or more codes to describe a procedurewhen a single, more comprehensive, code exists thataccurately describes the procedure

Global service periodviolations

Billing for a major procedure—such as surgery—andrelated procedures, when the fee for the major procedurealready includes the fee for related procedures providedduring a predefined time period (the global serviceperiod)

Duplicate procedures Billing for the same procedure twice although it was onlyprovided once

Unnecessary assistantsurgeon

Billing for an assistant surgeon when an assistant was notwarranted

Unbundling is a common type of abuse. Figure 1 illustrates howunbundling can lead to overpayment for an electrocardiogram. Using thisillustration, a provider would be overpaid if HCFA paid for both thecomprehensive service (93224) and one or more of its component parts(93225, 93226, or 93227). Overpayment occurs because the fee forperforming the comprehensive service already includes the value of thecomponent parts of the service. A provider would also be overpaid if HCFA

paid for all three individual components instead of the less expensivecomprehensive procedure, an unbundling practice called fragmentation.

4Many providers use commercial firms, called billing services, to prepare their claims.


B-261034

Figure 1: Structure of CPT Coding Scheme for Electrocardiogram

9322424 hour monitoring, including recording,scanning analysis with report,physician review and interpretation

93225recording

93226scanning analysis

with report

93227physician review and interpretation

This illustration can also be used to describe duplicate procedures. Aprovider would be overpaid if HCFA paid twice for this service on the sameday because the fee for the service covers a 24-hour period.

Computer Systems BeingUsed to Detect Abuses

Due to the large number of claims processed by Medicare carriers—about500 million claims for physician services and supplies in 1993—and thecomplexity of the coding system and payment policies, it is not feasible forcarrier staff to detect code manipulation by manually examining claims.To implement controls to prevent these abuses, HCFA has directed itscarriers to develop computer programs that (1) detect each type of abuseand (2) automatically adjust the payment. HCFA also provides carriers withthe specific code combinations that should not be accepted and directscarriers to incorporate the list in their computer systems.


B-261034

Because insurers have found it difficult to develop and maintain thespecialized capabilities required to detect code manipulation on their own,commercial firms have developed and now market systems that focus ondetecting this type of abuse. The complex analysis needed to quickly andaccurately (1) detect the numerous code combinations that could result inoverpayment, and (2) calculate the proper payment, requires sophisticatedcomputer programs.

Scope andMethodology

To determine whether commercially available codemanipulation-detection systems would save money, we conducted acontrolled test by having four commercial firms reprocess statisticallyvalid samples of over 200,000 claims. Each sample included claims forabout 24,000 beneficiaries that had been paid by Medicare during the first9 months of 1993, the most recent period for which data were available atthe time of our review. We controlled the test by ensuring that eachsystem’s capabilities were limited to detecting billing code abuses usingCPT codes that were valid in 1993. The systems did not, however, exactlymatch HCFA’s current code manipulation-detection rules because wewanted to compare Medicare to private industry practices. We alsoverified the test results by independently reviewing a random sample ofclaims each firm identified as having been overpaid. We confirmed that theadjustment made to each claim followed the appropriate system rule andthat the rule was supported by medical documentation. The scope of ourtest was limited to the $36 billion portion of the program that covers thecost of physician services and supplies.

To evaluate whether HCFA’s current development approach would matchcommercial system savings, we interviewed responsible HCFA officials andreviewed documents describing HCFA’s approach, scope, and methodology.We also reviewed documents describing HCFA’s preliminary results. Wecompared these preliminary results to existing commercial capabilities. Toassess the cost-effectiveness of commercial systems, we interviewedcommercial firm officials who provided cost estimates. We validated thereasonableness of each estimate by comparing it to the cost estimatedeveloped by a federal agency that recently decided to implement acommercial system. We also obtained oral comments on a draft of thisreport from the Deputy Directors of HCFA’s Bureau of Program Operationsand Bureau of Policy Development. Their views are summarized in theAgency Comments and Our Evaluation section of this report. Our workwas performed at HCFA headquarters in Baltimore, Md.; various Medicarecarriers; and offices of the four commercial firms from February 1994


B-261034

through April 1995 in accordance with generally accepted governmentauditing standards. Appendix I includes a detailed discussion of our scopeand methodology.

Commercial SystemsCould Save Over $600 Million a Year

HCFA could save over $600 million annually by using commercial systemsto detect code manipulation. Also, beneficiaries would save over$140 million a year. Although losses are substantial, less than 10 percent ofthe providers in our sample had one or more miscoded claims. Unbundlingand global service period violations made up 93 percent of the potentialsavings. According to several private and public insurers who usecommercial systems, our overall savings estimate is comparable to thesavings they have actually achieved with commercial systems.

Test Results Based on a controlled test conducted by four firms, commercial systemscould have reduced costs for physician services and supplies, on average,by about $452 million during the first 9 months of 1993, or about 1.8percent of outlays for those services.5 Extrapolating from those results,figure 2 shows that HCFA could have saved about $603 million in 1993 andabout $640 million in 1994.6 Appendix II identifies the participating firms.

5The number of claims included in our sample allows us to be 95 percent confident that actual savingswould have been within 5 percent of our estimate.

6We believe that our calendar year estimates reasonably approximate the extent of losses thatoccurred because HCFA did not significantly strengthen its controls during this time. We alsocompared 11 different claim characteristics, by carrier, to ensure that claims processed during the last3 months of 1993 had the same characteristics as the claims in our sample. We found no significantdifferences. This allows us to be confident that seasonal changes, such as (1) a possible shift in thebeneficiary population to the South or (2) changes in the types of medical services provided during thelast quarter of the year, would not affect the extent of abuse.


B-261034

Figure 2: Estimated Average Savings for Medicare Program

1993 (9 mo) 1993 (12 mo) 1994 (12 mo)$0

$100

$200

$300

$400

$500

$600

$700

$800

$452

$603$640

Savings

in millions

The savings estimates for the four firms were reasonably consistent,ranging between 1.4 and 2.2 percent of outlays. Medicare beneficiarieswould have saved $100 million during the first 9 months of 1993, whichextrapolates to $134 million in 1993 and $142 million in 1994.


B-261034

Savings would vary from our estimate for several reasons. First, savingscould be diluted somewhat by the results of provider appeals to havepayment reductions reconsidered. For example, some adjustments will bedue to coding errors which, when corrected, would result in payment. Onthe other hand, it is likely that commercial systems would generate moresavings than identified through the narrowly defined scope of our test. Toensure that savings were not overstated, we did not test some features ofcommercial system that are designed to generate savings. These featuresinclude (1) ensuring that procedures are appropriate to the beneficiary’sage and sex and (2) analyzing historical claims to identify patterns ofcoding abuse.

Although the potential savings are large, 92 percent of the providers in oursample billed correctly. Only 4 percent of the claims reviewed by the fourcommercial firms required adjustment. As shown in figure 3, fewer thanone in 12 providers had one or more claims adjusted by the commercialsystems. This is an important fact because, since most providers billcorrectly, most would not be affected by better controls to identify theseabuses.


B-261034

Figure 3: Percentage of Providers With Appropriately Coded and Miscoded Claims

8%

92%

Miscoded Claims

Appropriately Coded Claims

Types of Abuse Detected The commercial systems found abuse in each of the four categories. Twocategories, unbundling and global service period violations, accounted for93 percent of the savings in the claims sample. Figure 4 shows theproportion of savings in sampled claims by abuse type from two of thecommercial firms. These were the only firms that categorized savings byabuse type.


B-261034

Figure 4: Percentage of Savings by Type of Abuse

69%

24%

6%

1%

Unbundling

Global ServicePeriod Violations

DuplicateProcedures

Unnecessary Assistant Surgeon

The following examples, drawn from the sampled claims, illustrate (1) thenumerous and complex ways that procedure codes can be manipulated toincrease reimbursement and (2) the difficulty involved in detecting abuse.To protect against these abuses, computer systems must quickly comparemillions of possible code combinations that can be abused.

Unbundling includes several related abuses. Simple unbundling occurswhen a provider charges a comprehensive code as well as one or morecomponent codes. Because thousands of comprehensive codes exist with


B-261034

one or more component codes, numerous combinations of comprehensiveand component codes can be submitted on a claim. To identifyunbundling, the computer must be able to determine whether each codesubmitted is a component of one or more comprehensive codes. Figure 5shows how an x-ray examination was unbundled.

Figure 5: Example of Unbundling

A physician was paid for two x-ray exams of the abdominal region on the same date of service. According to CPT code descriptions, the x-ray of the upper gastrointestinal tract includes the x-ray of the abdomen.

CPT HCFA HCFA ShouldCode Procedure Allo wed Have Allowed

74000 x-ray exam of abdomen $23 $ 074241 x-ray exam of upper GI tract $75 $75

Total: $98 $75 Savings: $23

Allowed Should Allow$0

$20

$40

$60

$80

$100


B-261034

Fragmentation is a more complex and difficult-to-detect form ofunbundling. In this case, the provider bills for several component codesinstead of the more comprehensive code, which is normally less expensivethan the sum of the individual components. These abuses are difficult todetect because the computer must be able to recognize whichcombinations of component procedures equal a comprehensive procedureand then substitute a new code that was not included on the originalclaim. Figure 6 shows how an x-ray examination was fragmented.


B-261034

Figure 6: Example of Fragmentation

A physician was paid for interpreting two x-rays of the pelvis and two x-rays of the hip. According to CPT code descriptions, there is a more comprehensive CPT code -- 73520 -- that describes the four separate procedures as one.

Code HCFA HCFA ShouldSubmitted Procedure All owed Have Allo wed72170-26 x-ray of pelvis, 1 view $16 $ 0

(2 charges @ $8)73500-26 x-ray of hips, 1 view $16 $ 0

(2 charges @ $8)

Total: $32 $ 0Code Added:73520-26 x-rays of hips & pelvis, $16

2 views of each Savings: $16


$5

$10

$15

$20

$25

$30

$35

Mutually exclusive procedures—another form of unbundling—are thosethat are either impossible to perform together or, by accepted clinicalpractice standards, should not be performed at the same time. There are,however, caveats since, in some cases, a physician may try one approachand in mid-operation decide on another approach. The accepted paymentpractice in such circumstances is to pay for the more clinically intenseprocedure, not for both. To detect these abuses, the computer must be


B-261034

able to recognize which combinations of procedures either (1) should notbe performed together or (2) represent alternative approaches to deal withthe same problem. Figure 7 shows mutually exclusive laboratory tests.

Figure 7: Example of Mutually Exclusive Procedures

A physician was paid for two different antibiotic sensitivity tests which use different methods to achieve the same objective -- determining how effective an antibiotic is in treating the patient's bacterial infection. These two procedures, therefore, are considered to be mutually exclusive of each other. The physician billed this way eight different times.

CPT HCFA HCFA ShouldCode Proc edure All owed Have Allo wed

87181 antibiotic sensitivity; agar $484 $484 diffusion (3 charges @ $88, 5 @ $44)

87184 antibiotic sensitivity; disk $118 $ 0 (3 charges @ $21, 5 @ $11)



$100

$200

$300

$400

$500

$600


B-261034

Global service period violations are possible because the fee for mostsurgery includes all related services for a set number of days before andafter the surgery. Detecting these abuses can be difficult because thecomputer must be able to determine which services are related to thesurgery and which are not. Figure 8 shows a global service periodviolation.


B-261034

Figure 8: Example of Global Service Period Violation

A surgeon was paid for an office visit the day before a major surgical procedure. This visit should not have been paid because the fee for the surgery includes related services provided on the day before the surgery.

CPT HCFA HCFA ShouldCode Procedure All owed Have Allo wed

29881 knee arthroscopy $372 $37299213 office visit $ 32 $ 0

Total: $404 $372 Savings: $ 32


$100

$200

$300

$400

The difficulty in detecting global surgery violations is compounded whenservices are rendered by more than one provider. HCFA payment policyallows the fee to be divided, but does not allow the total payment toexceed the global fee. Figure 9, shows a case in which, because anassistant surgeon was involved, the computer must keep track of charges


B-261034

being made by the assistant to prevent overpayment. In addition to thepayments made to the assistant, a surgeon who performed the operationwas paid $1,219, which includes the value of related services during theglobal period.

Figure 9: Example of a Global Service Period Violation

An assistant surgeon was paid for open treatment of a broken femur which calls for a 90-day global service period. There were 20 inappropriate payments within the 90 days. Charges were submitted on six different claim forms.

CPT HCFA HCFA ShouldCode Procedure Allo wed Have Allo wed

27507-80 open treatment of broken femur $195 $19599212 office visit $ 21 $ 099223 initial hospital care $186 $ 0

(2 charges @ $93)99231 subsequent hospital care $243 $ 0

(9 charges @ $27)99232 subsequent hospital care $ 36 $ 099238 hospital discharge mgmt $ 94 $ 0

(2 charges @ $47)99311 nursing facility care $140 $ 0

(5 charges @ $28) Total: $915 $195

Savings: $720


$100

$200

$300

$400

$500

$600

$700

$800

$900


B-261034

A HCFA official told us that this example does not reflect existing HCFA

payment policy. According to this official, when an assistant surgeon isinvolved, HCFA allows the assistant to be paid for services that wouldnormally be included in global fee. However, even if HCFA policy allowsthese payments, the key point illustrated above is that HCFA is losingmoney by not enforcing global service fee periods for assistant surgeons,as is done in the private sector.

Duplicate procedures also exist in several forms, some of which can bedifficult to detect. Simple or exact duplicate procedures involves chargingfor the same procedure twice when it was only provided once. Evensimple duplicate procedures are not always easy to detect because it issometimes appropriate to pay more than once for the same service on asingle day. Therefore, the computer must be able to distinguish betweencodes that should and those that should not be paid for more than once ina single day. Figure 10 shows duplicate hospital care services.


B-261034

Figure 10: Example of Duplicate Procedures—Same Physician, Same Day, Same Site of Care

CPT HCFA HCFA ShouldCode Procedure Allo wed Have Allo wed99231 subsequent hospital care $ 59 $ 099231 subsequent hospital care $ 92 $ 92

Total: $151 $ 92 Savings: $ 59

A physician was paid twice for the same procedure -- subsequent hospital care. One claim covered 3 days and the second covered 2 days. The physician was entitled to receive payment for 3 days of care.


$50

$100

$150

Similarly, some procedures cover all services rendered regardless ofwhere the services were provided. Figure 11 illustrates an example ofduplicate procedures by charging for the same service provided threetimes at three different sites of care.


B-261034

Figure 11: Example of Duplicate Procedures—Same Physician, Same Day, Different Site of Care

CPT HCFA HCFA ShouldCode Procedure Allo wed Have Allo wed99214 office visit $ 53 $ 099223 initial hospital care $ 73 $ 099285 emergency department visit $110 $110


A physician was paid for 3 separate encounters with a patient for the same day of service for the same condition -- atrial fibrillation -- irregular contractions of the heart. Medicare allows a physician to only be paid for one visit per date of service if all encounters are for the same or related condition, except for critical care services. The charges were billed on three separate claim forms.


$40

$80

$120

$160

$200

$240

Private and Public InsurersConfirm CommercialSystem Savings

Commercial systems are widely used by private and public insurers.Officials we surveyed from both private and public insurers were satisfiedwith the benefits—both monetary and nonmonetary—generated for theircompanies by using commercial systems to detect code manipulation. Allof the officials also said that Medicare would benefit from usingcommercial systems.


B-261034

Almost 200 private insurers now use commercial systems to detect codemanipulation, including 13 of the 20 largest. In addition, several publicinsurers, such as state Medicaid agencies and Medicare contractors whoprovide services to beneficiaries enrolled in managed care plans usecommercial systems.7 The Department of Defenses’ Civilian Health andMedical Program of the Uniformed Services (CHAMPUS), which provideshealth insurance to dependents of military personnel, has also contractedto use a commercial system. Although 16 of the 32 Medicare carriers usethese systems to process claims for their private businesses, none uses acommercial system for its Medicare claims because HCFA directs them toimplement HCFA-developed controls.

We contacted 11 officials from private and public insurers that usecommercial systems. All of the officials stated they realized substantialsavings, although the benefits varied according to how each insurermodified the system and how each estimated savings. Six insurers statedthat their savings ranged from 1 to 2 percent of claims payments.8 TheCHAMPUS program, which generally follows Medicare payment restrictions,recently had one commercial firm test a sample of claims. The firmidentified potential savings totaling about 2 percent—similar to ourestimate of potential Medicare savings.

The officials also cited other benefits of using commercial systems. Nineofficials stated that commercial systems provided a clinically soundmethod for reviewing claims to detect code manipulation. That is, becausethe systems were developed with the support of physicians, codingdeterminations are closely tied to CPT code descriptions, and the inputfrom practicing physicians prevents the systems from denying claims forstrictly administrative reasons that do not make sense in patienttreatment. Two officials added that the commercial firms provide goodcustomer service and support in explaining coding adjustments toproviders. One noted that standardized explanations helped providersunderstand why code determinations were made, reducing the number ofappeals. Four officials cited the ability to easily modify the system to fittheir unique requirements as another benefit. Four officials saidcommercial systems also provided more consistent application of rules byeliminating human intervention and judgment.

7Medicare contracts separately for services to beneficiaries who are enrolled in managed care plans.These “risk” contractors agree to provide care to beneficiaries at a fixed fee. Several risk contractorsuse commercial code manipulation detection systems to control their costs.

8The remaining insurers’ estimates were not useful as comparisons because they involved estimates ofannual monetary savings or of the number or percentage of claims that were adjusted.


B-261034

HCFA’s DevelopmentApproach Will NotYield ComparableSavings

HCFA’s internal effort to better detect these abuses will not matchcommercial systems’ capabilities or savings. A primary reason is that HCFA

is not addressing the types of abuse that accounted for about one-third ofthe savings identified by commercial firms. HCFA also will not matchcommercial systems’ ability to detect unbundling. Commercial firms arebetter able to develop code manipulation-detection capabilities thanindividual insurers because they profit by excelling in theirspecialty—helping insurers detect billing abuses.

Scope of HCFA’s InitiativeLimits Potential Benefits

In August 1994, HCFA awarded a contract to strengthen its ability to identifyunbundling. HCFA’s contractor recently identified about 40,000 codes thatshould be denied when submitted with a another code. After review bymedical societies and final approval by HCFA, the new code combinationswill be incorporated in carrier claims processing computer systems by theend of this October. However, HCFA’s contract does not address two othertypes of abuse subject to significant losses. Our test results show thatsignificant amounts of global service period violations and duplicatebilling are not detected by HCFA carriers. These abuses accounted for about30 percent of the losses identified in the sample of claims tested by thetwo commercial firms that categorized savings by type of abuse.

Further, HCFA’s contract will not fully correct unbundling deficiencies. Incontrast to the 40,000 inappropriate code combinations identified byHCFA’s contractor, commercial systems are designed to analyze millions ofpotential combinations of component and comprehensive codes. Toestimate the extent to which HCFA’s approach would correct theunbundling deficiencies identified by commercial systems, we comparedHCFA’s proposed code combinations to faulty claims identified by onecommercial firm. First, we identified all inappropriate code combinationsdetected for a sample of 50 beneficiaries. We then compared theseproblem codes to HCFA’s proposed new code combinations. As shown infigure 12, HCFA’s proposed improvements would not have identified anyglobal surgery or duplicate claims. In addition, of 57 unbundled codesidentified by the commercial firm, HCFA’s proposal would have identifiedonly 13.


B-261034

Figure 12: Proportion of Abuses Correctable Under HCFA Effort

AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA








AAAAAAAAAAAAAAAAAAAAAAAAAA

35%

7%

45%

13%

Duplicate Procedures-- will not detect

Global Service Period Violations-- will not detect

Unbundling-- will not detect

Unbundling-- will detect

Commercial Firms Have anAdvantage

Because commercial firms focus on developing systems to detect codemanipulation, and do so as a business concern in a competitive market,they are better equipped to develop effective capabilities than areindividual insurers. Commercial firms invest significant full-time resourcesto identify the relationships among numerous codes and codecombinations that are subject to abuse. Commercial firms have multiplephysicians on staff and a network of board-certified consulting physiciansin specialty areas to analyze all codes and code combinations. Commercialfirms also employ computer professionals to develop efficient systems todetect code manipulation. In contrast, HCFA has invested limited resourcesto identify the numerous codes that can be abused. HCFA’s contract calledfor a single physician and limited support staff to identify newinappropriate code combinations. The contract does not call for anyactivity to improve computer system capabilities.


B-261034

Commercial firms are also better able to maintain up-to-date capabilities,compared with HCFA’s development schedule. Maintaining up-to-dateabuse-detection capabilities is difficult because a substantial number ofprocedure codes change each year. The firms received the revised CPT

code manual in October 1994 and all provided their customers with systemchanges that incorporated new code combinations this spring. In contrast,the code combinations identified by HCFA’s contractor will not beimplemented by Medicare until October 1995—6 months behind thecommercial systems. Unless HCFA changes its contract schedule,subsequent changes to maintain current code manipulation-detectioncapabilities will be similarly delayed.

Commercial firms are also better positioned to develop productimprovements to counter new types of abuse because they interact with alarge number of insurance clients who demand new capabilities to controlnewly detected abuses. To remain competitive, commercial firms have anincentive to respond quickly. They do so by issuing annual productimprovements. According to one firm, a customer recently identified apotential new abuse: an increasing number of childbirth claims includedcharges for such physical therapy services as whirlpool baths andmassages. Because childbirth is covered under a global service period,related services should not be charged separately. Although the firm’ssystem checked services related to childbirth, it did not check for physicaltherapy services. The firm’s officials explained that the customer addedphysical therapy services to the system’s childbirth checks. The firm isnow analyzing claims data and medical literature related to physicaltherapy and childbirth to determine whether similar checks should beadded to its standard system.

Commercial SystemsAre Cost-Effective

Potential savings of over $600 million a year, compared with acquisitioncosts of about $20 million, make commercial systems a highlycost-effective investment. The four firms that participated in our testestimate that the cost to implement and operate a commercial system for 1year would range from $10 million to $20 million at all 32 Medicarecarriers. The actual cost would be subject to formal bids and negotiationswith interested firms. One reason for the wide range in estimates isuncertainty about the technical requirements to implement a commercialsystem with existing carrier computer systems. The $20-million estimateanticipates unknown problems in attempting to implement a commercialsystem with the seven different claims processing systems currently usedby Medicare carriers.


B-261034

The experience of the CHAMPUS program lends credence to commercialfirm estimates and the caution that unanticipated problems could occur.CHAMPUS uses five contractors to process its claims. CHAMPUS officials toldus that they estimate the annual cost to license and implement acommercial system at all five contractor locations will be under $2 million.They also noted that careful planning is appropriate becauseimplementation difficulties can occur. The CHAMPUS program encounteredunanticipated delays implementing a commercial system. According toprogram officials, the agency needed to change existing claims processingsystems and the commercial system more than expected. These officialsstated, however, that implementation delays could be avoided by fullyanalyzing the required changes when evaluating commercial systems.

Conclusions Fraud, waste, and abuse are problems faced by all health insurers. HCFA, asthe agency responsible for administering the nation’s largest insuranceprogram, could have been a leader in implementing effective paymentcontrols to prevent losses to billing abuse. However, HCFA has not keptpace with private industry’s use of advanced information technology todetect code manipulation, one common form of abuse. As a result, overhalf a billion dollars is being wasted each year. HCFA’s internal efforts todevelop code manipulation-detection capabilities are limited and will notfully stem losses from these abuses.

HCFA could benefit from the experiences of private and other publicinsurers who have turned to commercial systems to enhance their abilityto control costs by avoiding payments for faulty claims. Such systemsprovide a more comprehensive ability to protect Medicare funds. In an eraof reinventing government initiatives, existing agency perceptions ofopportunities and limitations must be reexamined; bold ways to betteraccomplish missions and protect government resources can be identified.Acquiring commercial systems represents such bold thinking, andprovides an efficient and cost-effective way to reduce Medicare programlosses substantially.

Recommendations To better protect Medicare funds from losses due to code manipulation,we recommend that the Secretary, HHS, direct the Administrator of HCFA torequire Medicare carriers to use a commercial system to detect codemanipulation when processing Medicare claims for physician services andsupplies.


B-261034

Agency Commentsand Our Evaluation

Senior HCFA officials provided oral comments to our draft report. Theseofficials stated that HCFA supports the use of modern informationtechnology to strengthen payment controls. They also stated that HCFA willfully analyze the feasibility of using commercially available codemanipulation-detection software to process Medicare claims.

The officials cautioned, however, that HCFA has a responsibility as a publicagency to resolve three important issues before requiring carriers toimplement commercial technology. First, to ensure that commercialsystems adjust claims appropriately, HCFA needs assurance thatcommercial system rules match or can be modified to match Medicarepayment policies. Second, to ensure that physicians and other affectedparties have an opportunity to provide comments on Medicare policies,HCFA needs to determine the extent to which commercial firms would bewilling to disclose information about their systems. Third, HCFA needs toanalyze the cost and technical feasibility of implementing commercialsystems with existing carrier claims processing systems. These officialsnoted that HCFA has scheduled briefings with each firm to begin addressingthese issues.

We believe these issues can be resolved. First, commercial firm officialstold us that their systems are designed to be easily customized toimplement different payment policies. This would also give HCFA theopportunity to reassess its current payment policies when analyzingcommercial system capabilities. As noted in this report, a HCFA officialindicated that a global surgery period overpayment detected by onecommercial firm would not be prevented under current HCFA paymentpolicies. Second, although commercial firm officials consider the details oftheir computer systems to be proprietary, and not publicly releasable, theytold us that within certain parameters, HCFA could obtain input fromaffected parties. Accordingly, HCFA could continue to release Medicarepayment policies and detailed examples of the types of code combinationsthat are inapproprite based on the policies. Third, as pointed out in thisreport, the estimated cost to implement commercial systems is from$10 million to $20 million. Regarding technical feasibility, commercial firmofficials told us that their systems are designed to operate with a widevariety of claims processing systems and to be easily installed. Thiscapability is illustrated by the fact that commercial systems are widelyused by private insurers.

HCFA officials also expressed concern that we did not fairly portray HCFA

efforts to prevent billing abuse, including code manipulation. They stated


B-261034

that HCFA has made significant progress in deterring abusive billing, citingefforts to implement physician payment reforms, including regulations tostandardize payment rules and strengthen controls to prevent globalsurgery period violations. While we applaud these efforts, our test resultsshow that commercial systems provide an opportunity to furtherstrengthen HCFA’s ability to deter these abuses.

We are sending copies of this report to the Secretary of HHS, theAdministrator of HCFA, the Office of Management and Budget, andMedicare carriers. Copies also will be made available to others uponrequest. This report was prepared under the direction of Patricia T. Taylor,Associate Director, Information Resources Management/Health,Education, and Human Services. If you have any questions regarding thisreport, you can contact me at (202) 512-6252 or her at (202) 512-5539.Other major contributors are listed in appendix III.

Frank W. ReillyDirector, Information Resources Management/ Health, Education, and Human Services



Contents

Letter 1

Appendix I Scope andMethodology

32

Appendix II Commercial FirmsThat Participated inThis Review

36

Appendix III Major Contributors toThis Report

37

Table Table 1: Categories of Abuse 4

Figures Figure 1: Structure of CPT Coding Scheme for Electrocardiogram 5Figure 2: Estimated Average Savings for Medicare Program 8Figure 3: Percentage of Providers With Appropriately Coded and

Miscoded Claims10

Figure 4: Percentage of Savings by Type of Abuse 11Figure 5: Example of Unbundling 12Figure 6: Example of Fragmentation 14Figure 7: Example of Mutually Exclusive Procedures 15Figure 8: Example of Global Service Period Violation 17Figure 9: Example of a Global Service Period Violation 18Figure 10: Example of Duplicate Procedures—Same Physician,

Same Day, Same Site of Care20

Figure 11: Example of Duplicate Procedures—Same Physician,Same Day, Different Site of Care

21

Figure 12: Proportion of Abuses Correctable Under HCFA Effort 24


Contents

Abbreviations

AIMD Accounting and Information Management DivisionCHAMPUS Civilian Health and Medical Program of the Uniformed

ServicesCPT common procedural terminologyGAO General Accounting OfficeHCFA Health Care Financing AdministrationHHS Department of Health and Human ServicesSAF standard analytical file


Appendix I

Scope and Methodology

To determine whether commercial systems would save money, weconducted a controlled test by having four commercial firms reprocess asample of claims that Medicare paid during the first 9 months of 1993—themost recent time period for which data were available at the time of ourreview. Although these billing abuses affect the entire $56 billion part Bportion of the Medicare program, the scope of our test was limited toclaims for physician services and supplies, which cost $36 billion in 1994.We did not test other categories of Medicare part B claims because(1) HCFA’s claims history file did not maintain the information needed todetect billing abuse on outpatient claims and (2) independent laboratory,prepaid group practice, and home health services account for a relativelysmall portion of part B costs.

Four Commercial FirmsAgreed to DemonstrateTheir Systems’ Capabilities

To determine whether commercial systems are more capable of detectingabuse than systems Medicare uses, we arranged for a controlled test of thecapabilities of four off-the-shelf commercial systems that insurers use todetect abuse. To identify which commercial firms market these systems,we (1) reviewed literature describing computer products used in theclaims-processing industry, (2) contacted the HHS Office of InspectorGeneral, which had analyzed commercial capabilities, and a Department ofDefense health insurance agency, which was considering a commercialsystem, (3) talked with exhibitors attending a national health careantifraud conference, and (4) contacted companies marketingabuse-detection systems to determine if they would be willing toparticipate in our evaluation.

All four commercial firms we identified agreed to participate. We heldseveral discussions with each company to determine its product’scapability and market penetration and arrange the terms of participation.Two issues were central to these discussions. First, the companies wantedassurances, which we provided, that we would not disclose proprietaryinformation about their systems. Second, we designed the test to avoid adirect or implied comparison of company capabilities because ourobjectives did not include identifying which system would best meetHCFA’s needs. We took several steps to avoid such a comparison, includingproviding different claims samples to each company, controlling the editsthat each company applied, and using average results, rather than eachcompany’s results in our report. We documented the study requirementsand ground rules in a memorandum of understanding between GAO andeach firm. Also, in October 1994, we identified a company that recently


Appendix I


began marketing a billing abuse detection product. This firm briefed us onits system capabilities but did not participate in the study.

We Obtained a ValidSample of Medicare Claims

To obtain a valid sample of Medicare claims, we reviewed documentsdescribing the contents of HCFA’s Medicare databases and held severaldiscussions with responsible HCFA officials about the content andreliability of the data. We then selected the data elements required for theanalysis and confirmed that the elements corresponded to the data neededby each company.

We selected HCFA’s 5-percent standard analytical file (SAF) as theappropriate source for the sample. The 5-percent SAF contains final actionclaims—reviewed and validated—for a random sample of 5 percent ofMedicare beneficiaries. These claim records are obtained directly fromHCFA’s common working file system, the system that authorizes claimspayments. The 5-percent SAF is also used extensively by HCFA and publicpolicy researchers, and is the primary source of data about the Medicareprogram. HCFA documents cited controls and quality assurance testing toensure data reliability. To further verify the reliability of the HCFA data, weanalyzed the controls over the process used to convert data from thecommon working file to the 5-percent SAF, and reviewed the results of HCFA

quality assurance assessments.

The data were generally reliable for our purposes, with one exception. Theamounts contained in the common working file and 5-percent SAF as beingpaid did not always reflect the amount that was actually paid. Thisdiscrepancy exists because, in some cases, the common working filedirected carriers to recalculate the paid amount but did not record theadjustment. This problem has been subsequently corrected. The paidamounts are important to this analysis because they represent the amountof federal outlays for the Medicare program. After discussing this issuewith staff from HCFA’s Office of the Actuary, we decided, and HCFA agreed,that we could closely approximate the amount of federal outlays byreducing the allowed amount1 by 22 percent which, according to HCFA

actuaries, is the amount beneficiaries actually paid in coinsurance anddeductibles.

We confirmed that HCFA’s method of selecting the beneficiary sample forthe 5-percent SAF was statistically valid, obtained a list of all beneficiaries

1The allowed amount is the amount HCFA authorizes the physician to collect. Federaloutlays—reflected in the paid amount—are calculated by deducting beneficiary deductible andcoinsurance obligations from the allowed amount.


Appendix I


included in the 5-percent SAF, and selected a statistically valid randomsample of beneficiaries. We asked HCFA to extract all applicable claimsprocessed during the first 9 months of 1993 for the sampled beneficiaries,the most recent data available at the time of our review.

To convert the data into a format that each company could use, weworked directly with each company’s technical representatives tounderstand their data record layout requirements and develop theprograms necessary to convert HCFA’s data to the required formats. Wethen divided the claims and converted each group into the format neededby each company.

We Controlled the Test andVerified the Results

To ensure that the test was limited to identifying instances of codemanipulation, we reviewed each company’s user manuals, systemmanuals, and payment rules to understand the basis for each type of ruleand the sequence with which the system executed its analysis. We thendiscussed the research that went into determining each rule type toensure, as far as possible, that the test would be limited to clear-cutinstances of code manipulation that did not need manual intervention inorder for a decision to be made. We divided the rule types into threecategories: checks that identify inappropriate payments; other checks thatcould lead to savings but either involve manual review or could reflectdata entry errors; and checks that were outside the scope of our review.Our savings estimates were limited to the first category—inappropriatepayments that could be automatically detected and adjusted on the basisof the data contained on the claims. We also controlled the test byensuring that each system’s capabilities were limited to detecting abusesusing CPT codes that were valid in 1993. Because we wanted to compareMedicare to private industry practices, the systems were not customizedto reflect HCFA payment rules which, in some cases, differ from those ofprivate insurers.

To verify the accuracy of the companies’ analyses, we selected andreviewed a random sample of claims that were adjusted by each company.We compared the firms’ actions with CPT code descriptions and paymentrules used by the system. We met with company representatives to revieweach claim and verify that the adjustment made was based on adocumented rule, supported by medical analysis, and processedaccurately by the system.


Appendix I


We Evaluated HCFA’sEffort To Develop BillingAbuse-DetectionCapabilities

To evaluate whether HCFA’s current development approach would matchcommercial system savings, we interviewed responsible HCFA officials andreviewed documents describing the approach, scope, and methodologybeing followed. We reviewed the contract HCFA awarded to defineadditional unbundled code combinations to determine its scope,methodology, resource requirements, and schedule. We compared HCFA’sapproach with that used by commercial firms. We also reviewed twoproducts of the contract that described the improvements expected. Thedraft Medicare unbundling policy and new unbundled code combinationsprovided a basis to estimate the extent to which HCFA’s proposedimprovements would incorporate capabilities available in commercialsystems. We also reviewed the contractor’s analysis of existing Medicarecomputer system limitations and recommendations for near-term andlong-term improvements which explained why Medicare computersystems would not be able to match commercial system capabilities.

We Assessed theCost-Effectiveness ofCommercial Systems

To assess the cost-effectiveness of commercial systems, we interviewedcommercial firm officials who provided cost estimates. We validated thereasonableness of the estimate by interviewing officials from theDepartment of Defense’s Civilian Health and Medical Program of theUniformed Services (CHAMPUS), which provides health insurance todependents of military personnel. We compared it to the cost estimatedeveloped by a federal agency which recently decided to implement acommercial system.


Appendix II

Commercial Firms That Participated in ThisReview


Appendix III

Major Contributors to This Report

Accounting andInformationManagement Division,Washington, D.C.

David B. Alston, Assistant DirectorTheodore P. Alves, Assignment ManagerM. Yvonne Sanchez, Computer SpecialistYvette R. Banks, Technical AdviserMichael P. Fruitman, Communications AnalystTeresa L. Jones, Information Processing Specialist

Program Evaluationand MethodologyDivision, Washington, D.C.

Harry M. Conley III, Assistant Director

Cincinnati RegionalOffice

Kenneth B. Bibb, Evaluator-in-ChargeJulie A. Schneiberg, Computer SpecialistArthur D. Foreman, Technical Assistance Manager

Atlanta RegionalOffice

Amanda S. Cooksey, Staff Evaluator

San FranciscoRegional Office

Donald R. Hunts, Senior Evaluator

(511167) GAO/AIMD-95-135 Medicare Claims: Technology Could Save BillionsPage 37

Ordering Information

The first copy of each GAO report and testimony is free.

Additional copies are $2 each. Orders should be sent to the

following address, accompanied by a check or money order

made out to the Superintendent of Documents, when

necessary. Orders for 100 or more copies to be mailed to a

single address are discounted 25 percent.

Orders by mail:

U.S. General Accounting Office

P.O. Box 6015

Gaithersburg, MD 20884-6015

or visit:

Room 1100

700 4th St. NW (corner of 4th and G Sts. NW)

U.S. General Accounting Office

Washington, DC

Orders may also be placed by calling (202) 512-6000

or by using fax number (301) 258-4066, or TDD (301) 413-0006.

Each day, GAO issues a list of newly available reports and

testimony. To receive facsimile copies of the daily list or any

list from the past 30 days, please call (301) 258-4097 using a

touchtone phone. A recorded menu will provide information on

how to obtain these lists.

PRINTED ON RECYCLED PAPER

United StatesGeneral Accounting OfficeWashington, D.C. 20548-0001

Official BusinessPenalty for Private Use $300

Address Correction Requested

Bulk MailPostage & Fees Paid

GAOPermit No. G100

GAO/AIMD-95-135 Medicare Claims: Technology Could Save Billions

Date post:	01-Oct-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times